A model based on Markov processes was developed to compare the economic and epidemiologic effects of different strategies for control of paratuberculosis in a flock of sheep. The strategies have an annual periodicity and are: (I) no intervention; (II) testing and culling; (III) vaccination; (IV) testing and culling together with vaccination. The values of the variables were defined for the analysis of a typical case for a period of 10 years. The main analysis was performed using a theoretical serologic method with 95% sensitivity and specificity. The vaccine was considered to be 90% effective and to cause 20% cross-reactions. In addition, the results at different levels of sensitivity and specificity, vaccine efficacy and absence of cross-reactions to the vaccine were checked. The size of the flock (200) and the annual replacement rate by causes other than paratuberculosis (15%) were fixed on the basis of average figures for the Basque Country. The costs of each strategy were calculated on the basis of actual costs of each item and were used to estimate a benefit/cost ratio. Although eradication of the infection could be reached by any of the three methods within 10 years, in a situation of low prevalence the costs for methods with culling are higher than those of no intervention (benefit/cost ratio of 0.42 for culling alone and 0.19 for culling and vaccination). At high prevalence, these methods would become more profitable, but would still not cover the costs of the strategy (benefit/cost ratio of 0.86 and 0.48, respectively). The high costs and the difficulties of replacing the culled animals in the first years made these strategies very difficult to implement. Thus, the best strategy appeared to be to vaccinate the replacement ewe lambs. This strategy quickly reached a positive benefit/cost ratio at both levels of prevalence (2.19 and 7.10, respectively) and did not demand heavy initial investment. © 1993.